<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: smaddox</title><link>https://news.ycombinator.com/user?id=smaddox</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 10 Jun 2026 08:22:30 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=smaddox" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by smaddox in "How attention sinks keep language models stable"]]></title><description><![CDATA[
<p>> though we did not <i>delve</i> into the observation<p>Oh, the irony.</p>
]]></description><pubDate>Fri, 08 Aug 2025 23:40:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=44842811</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=44842811</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44842811</guid></item><item><title><![CDATA[New comment by smaddox in "Gold Is So Popular It's Making People Nervous"]]></title><description><![CDATA[
<p>Gold/M2SL(Billion USD) is currently around 0.12. In 1980, it peaked around 0.45. Monthly average since 1960 is 0.11. In late 2011 it peaked around 0.18.<p>Gold / Global M2 would be a better metric, but I haven't analyzed that yet.</p>
]]></description><pubDate>Sun, 27 Apr 2025 05:34:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=43809679</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=43809679</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43809679</guid></item><item><title><![CDATA[New comment by smaddox in "Cursor IDE support hallucinates lockout policy, causes user cancellations"]]></title><description><![CDATA[
<p>I switched to Windsurf.ai when cursor broke for me. Seems about the same but less buggy. Haven't used it in the last couple weeks, though, so YMMV.</p>
]]></description><pubDate>Tue, 15 Apr 2025 22:14:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=43699081</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=43699081</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43699081</guid></item><item><title><![CDATA[New comment by smaddox in "Electron band structure in germanium, my ass (2001)"]]></title><description><![CDATA[
<p>For those who are actually interested in this field, the proper way to measure this would be with a four point probe. You do need a constant current source and a high-impedence voltage meter, though.<p>Also, you don't need to solder wires to the sample. But if you want to measure the hall resistance of a thin film of a semiconductor, you can solder a glob of indium on to four corners of a 1 cm x 1 cm wafer, put it in a magnetic field, and then do basically the same measurement as four point probe, except not inline.</p>
]]></description><pubDate>Wed, 02 Apr 2025 04:11:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=43553610</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=43553610</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43553610</guid></item><item><title><![CDATA[New comment by smaddox in "Ask HN: Is anyone doing anything cool with tiny language models?"]]></title><description><![CDATA[
<p>You can train that size of a model on ~1 billion tokens in ~3 minutes on a rented 8xH100 80GB node (~$9/hr on Lambda Labs, RunPod io, etc.) using the NanoGPT speed run repo: <a href="https://github.com/KellerJordan/modded-nanogpt">https://github.com/KellerJordan/modded-nanogpt</a><p>For that short of a run, you'll spend more time waiting for the node to come up, downloading the dataset, and compiling the model, though.</p>
]]></description><pubDate>Tue, 21 Jan 2025 22:11:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=42785728</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42785728</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42785728</guid></item><item><title><![CDATA[New comment by smaddox in "Natrium 'advanced nuclear' power plant wins Wyoming permit"]]></title><description><![CDATA[
<p>> That application was submitted in March 2024 and is on track for approval in December 2026<p>Huh? Is this something where there's multiple incremental steps in the process, and that date is just the final approval stamp, or does it actually just take more than 1.5 years?</p>
]]></description><pubDate>Wed, 15 Jan 2025 20:24:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=42716408</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42716408</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42716408</guid></item><item><title><![CDATA[New comment by smaddox in "Byte Latent Transformer: Patches Scale Better Than Tokens"]]></title><description><![CDATA[
<p>Agree more levels seems like it could be beneficial. And another Meta paper published a day later shows how that might work: <a href="https://ai.meta.com/research/publications/large-concept-models-language-modeling-in-a-sentence-representation-space/" rel="nofollow">https://ai.meta.com/research/publications/large-concept-mode...</a></p>
]]></description><pubDate>Sun, 15 Dec 2024 08:06:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=42422235</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42422235</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42422235</guid></item><item><title><![CDATA[New comment by smaddox in "GM exits robotaxi market, will bring Cruise operations in house"]]></title><description><![CDATA[
<p>Oh, interesting. 120db is much better than what I thought possible with CMOS sensors. Thats competetive with the human eye.<p>Well, that definitely changes my opinion on how feasible/competitive camera-only autonomous driving can be.</p>
]]></description><pubDate>Thu, 12 Dec 2024 23:40:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=42404564</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42404564</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42404564</guid></item><item><title><![CDATA[New comment by smaddox in "GM exits robotaxi market, will bring Cruise operations in house"]]></title><description><![CDATA[
<p>Apparently they're using this CMOS sensor: <a href="https://www.onsemi.com/products/sensors/image-sensors/ar0136at" rel="nofollow">https://www.onsemi.com/products/sensors/image-sensors/ar0136...</a><p>It's not an event camera, so it's very much taking images, which are then being processed by computer vision algorithms.<p>Event cameras seem more viable than CMOS sensors for autonomous vehicle applications in the absence of LIDAR. CMOS dynamic range and response isn't as good as the human eye. LIDAR+CMOS is considerably better in many ways.</p>
]]></description><pubDate>Wed, 11 Dec 2024 04:24:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=42384780</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42384780</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42384780</guid></item><item><title><![CDATA[New comment by smaddox in "Intel’s board, and an example of when boards and short-termism fail"]]></title><description><![CDATA[
<p>> if you compare Intel's headcount with comparable companies (AMD, Nvidia), you can see Intel is really wasteful<p>AMD and NVIDIA are fabless. They are not comparable. It takes far more people to R&D a cutting-edge process node and run a dozen fabs 24/7 365.25 than it takes to design cutting edge chips.</p>
]]></description><pubDate>Fri, 06 Dec 2024 17:49:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=42342152</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42342152</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42342152</guid></item><item><title><![CDATA[New comment by smaddox in "Frosted Glass from Games to the Web: HTML glass UI inspired by AAA game dev"]]></title><description><![CDATA[
<p>If both sides are frosted, then you will have a similar effect as subsurface scattering.</p>
]]></description><pubDate>Sun, 24 Nov 2024 05:05:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=42226013</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42226013</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42226013</guid></item><item><title><![CDATA[New comment by smaddox in "/usr/bin/env -S uv run"]]></title><description><![CDATA[
<p>I've been using this for a few weeks now, and it's really handy. But I did learn the hard way that it fails if you don't have internet connection, even if you already have the venv cached.</p>
]]></description><pubDate>Thu, 21 Nov 2024 04:24:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=42201103</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=42201103</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42201103</guid></item><item><title><![CDATA[New comment by smaddox in "Breaking Bad: How Compilers Break Constant-Time~Implementations"]]></title><description><![CDATA[
<p>Not the OP, and not at all an expert in this area, but I was curious what the answer was, and from a bit of reading it seems like a potential reason that might not work is because volatile also prevents caching the value in a register. So if you want to keep the computation in a register, but you want to explicitly clear the value to zero before writing to it, it seems like C semantics are insufficient (at least without inline assembly).</p>
]]></description><pubDate>Wed, 23 Oct 2024 16:24:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=41926609</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41926609</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41926609</guid></item><item><title><![CDATA[New comment by smaddox in "First images from Euclid are in"]]></title><description><![CDATA[
<p>The slowdown is in the simulation, not in the internally perceived flow of time. The internally perceived flow of time would of course not change.<p>The problem is that an exponential slowdown at each level requires discounting the probability of being in a simulation by an exponential amount per level. So instead of being able to say that the weighted odds of being in a simulation are higher than the odds of <i>not</i> being in a simulation, you have to say that, at best, there's a small chance of being in a simulation.</p>
]]></description><pubDate>Wed, 23 Oct 2024 14:24:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=41925335</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41925335</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41925335</guid></item><item><title><![CDATA[New comment by smaddox in "First images from Euclid are in"]]></title><description><![CDATA[
<p>Photocopying has little to do with simulation of the physical world.<p>First of all, Bits != Q-bits. You can clone bits. You can't clone Q-bits: <a href="https://en.m.wikipedia.org/wiki/No-cloning_theorem" rel="nofollow">https://en.m.wikipedia.org/wiki/No-cloning_theorem</a><p>Second, photocopies are static. The physical world is not static.</p>
]]></description><pubDate>Tue, 22 Oct 2024 02:54:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=41910762</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41910762</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41910762</guid></item><item><title><![CDATA[New comment by smaddox in "First images from Euclid are in"]]></title><description><![CDATA[
<p>Exponential slowdowns at each level ruin this hypothesis.</p>
]]></description><pubDate>Tue, 22 Oct 2024 01:06:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=41910200</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41910200</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41910200</guid></item><item><title><![CDATA[New comment by smaddox in "I Went from Reading 40 Books a Year to Reading 0"]]></title><description><![CDATA[
<p>If you think novels, especially good ones, are "stories of made-up people that aren't intended to illustrate any particular idea", then you're missing quite a bit.</p>
]]></description><pubDate>Thu, 17 Oct 2024 16:04:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=41870922</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41870922</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41870922</guid></item><item><title><![CDATA[New comment by smaddox in "On the Nature of Time"]]></title><description><![CDATA[
<p>This model of physics does make some falsifiable predictions, and there are discussions about how to test them elsewhere.<p>Unlike string theory, this theory does not have any free variables to adjust. It's either true or it's false.<p>I, for one, find it to be trivially true. It fits every observation and is the only theory ever posed that doesn't have the "But why <i>those</i> initial conditions?" problem.</p>
]]></description><pubDate>Wed, 09 Oct 2024 18:55:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=41791372</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41791372</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41791372</guid></item><item><title><![CDATA[New comment by smaddox in "Silicon Valley, the new lobbying monster"]]></title><description><![CDATA[
<p>I can understand how the "up to an annual contribution limit" part of my comment could be confusing. I added that in an attempt to not be misleading / to circumvent spurious disagreement. I've moved it inside of parenthesis to de-emphasise it.<p>If you read my comment as a response to the parent, I believe it will make more sense.<p>---<p>CU opened the floodgates of anonymous, unlimited political contributions to "independent", but not really independent, organizations which can campaign on behalf of candidates.<p>The goal was/is to replace our democratic republic with an oligarchy. And we're now in the end game.</p>
]]></description><pubDate>Mon, 07 Oct 2024 19:34:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=41770034</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41770034</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41770034</guid></item><item><title><![CDATA[New comment by smaddox in "Silicon Valley, the new lobbying monster"]]></title><description><![CDATA[
<p>False dichotomy. Prior to CU, individuals could contribute to any candidate of their choosing (up to an annual contribution limit). Companies are compased of individuals. CU was twisted logic to reach a preconceived, corrupt outcome.</p>
]]></description><pubDate>Mon, 07 Oct 2024 18:06:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=41769002</link><dc:creator>smaddox</dc:creator><comments>https://news.ycombinator.com/item?id=41769002</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41769002</guid></item></channel></rss>